apskdemod

Amplitude phase shift keying (APSK) demodulation

Syntax

Z = apskdemod(Y,M,radii)

Z = apskdemod(Y,M,radii,phaseoffset)

Z = apskdemod(___,Name=Value)

Description

Z = apskdemod(Y,M,radii) performs APSK demodulation of the input signal Y, based on the specified number of constellation points per PSK ring, M, and the radius of each PSK ring, radii. For a description of APSK demodulation, see APSK Hard Demodulation and APSK Soft Demodulation.

Note

apskdemod specifically applies to multiple ring PSK constellations. For a single ring PSK constellation, use pskdemod.

Z = apskdemod(Y,M,radii,phaseoffset) specifies an initial phase offset for each PSK ring of the APSK modulated signal.

Z = apskdemod(___,Name=Value) specifies optional name-value arguments using any of the previous syntaxes. For example, apskdemod(Y,M,PlotConstellation=true) demodulates using constellation points per ring specified in input vector M and plots the constellation. Specify name-value arguments after all other input arguments.

Examples

Demodulate 16-APSK Signal

Demodulate a 16-APSK signal that has an unequal number of constellation points on each circle. Plot the received constellation.

Define vectors for modulation order and PSK ring radii. Generate random 16-ary data symbols.

M = [4 12];
radii = [1 2];
modOrder = sum(M);

x = randi([0 modOrder-1],1000,1);

Apply APSK modulation to the data.

txSig = apskmod(x,M,radii);

Pass the modulated signal through a noisy channel.

snr = 20; % dB
rxSig = awgn(txSig,snr,'measured');

Plot the transmitted (reference) signal points and the noisy received signal points.

plot(rxSig,'b*')
hold on
grid
plot(txSig,'r+')
xlim([-3 3])
ylim([-3 3])
xlabel('In-Phase')
ylabel('Quadrature')
legend('Received constellation','Reference constellation')

Demodulate the received signal and compare to the input data.

z = apskdemod(rxSig,M,radii);
isequal(x,z)

ans = logical
   0

Demodulate 64-APSK Custom Symbol Mapped Signal

Demodulate a 64-APSK signal with custom symbol mapping. Compute hard decision bit output and verify that the input matches the output.

Define vectors for modulation order and PSK ring radii. Generate 100 symbols of random bit input.

M = [8 12 16 28]; % 4-PSK circles
modOrder = sum(M);
radii = [0.5 1 1.3 2];
x = randi([0 1],100*log2(modOrder),1);

Create a custom symbol mapping vector of binary mapping.

cmap = 0:63; 

Modulate the data and plot the constellation.

y = apskmod(x,M,radii, ...
        SymbolMapping=cmap, ...
        InputType='bit', ...
        PlotConstellation=true);

Demodulate the received signal.

z = apskdemod(y,M,radii,SymbolMapping=cmap,OutputType='bit');

Verify that the demodulated signal is equal to the original data.

isequal(x,z)

ans = logical
   1

Soft Bit Demodulate 32-APSK Signal

Demodulate a 32-APSK signal and calculate soft bits.

Define vectors for modulation order and PSK ring radii. Generate 10000 symbols of random bit data.

M = [16 16];
modOrder = sum(M);
radii = [0.6 1.2];
numSym = 10000;
x = randi([0 1], numSym*log2(modOrder),1);

Generate a reference constellation. Create a constellation diagram object.

refAPSK = apskmod(0:modOrder-1,M,radii);
constDiagAPSK = comm.ConstellationDiagram( ...
    ReferenceConstellation=refAPSK, ...
    Title='Received Symbols', ...
    XLimits=[-2 2], ...
    YLimits=[-2 2]);

Modulate the data.

txSig = apskmod(x,M,radii,InputType='bit');
sigPow = var(txSig);

Pass the signal through a noisy channel.

snr = 15;
rxSig = awgn(txSig,snr,sigPow,'linear');

Plot the reference and received constellation symbols.

constDiagAPSK(rxSig)

Demodulate the signal and compute soft bits.

z = apskdemod(rxSig,M,radii, ...
    OutputType='approxllr', ...
    NoiseVariance=sigPow/snr);

Input Arguments

Y — APSK modulated signal

scalar | vector | matrix

APSK modulated signal, specified as a complex scalar, vector, or matrix. When Y is a matrix, each column is treated as an independent channel.

Data Types: double | single
Complex Number Support: Yes

M — Constellation points per PSK ring

integer vector

Constellation points per PSK ring, specified as an integer vector with more than one element. Vector elements indicate the number of constellation points in each PSK ring. The first element corresponds to the innermost circle, and so on, until the last element, that corresponds to the outermost circle. Element values must be multiples of four and sum(M) must be a power of two. The modulation order is the total number of points in the signal constellation and equals the sum of the vector elements, sum(M).

Example: [4 12 16] specifies a three PSK ring constellation with a modulation order of sum(M) = 32.

Data Types: double

radii — Radius per PSK ring

vector

Radius per PSK ring, specified as a vector with the same length as M. The first element corresponds to the innermost circle, and so on, until the last element, that corresponds to the outermost circle. The elements must be positive and arranged in increasing order.

Example: [0.5 1 2] defines constellation PSK ring radii. The inner ring has a radius of 0.5, the second ring has a radius of 1.0, and the outer ring has a radius of 2.0.

Data Types: double

phaseoffset — PSK ring phase offsets

[pi/M(1) pi/M(2) … pi/M(end)] (default) | scalar | vector

Phase offset of each PSK ring in radians, specified as a scalar or vector with the same length as M. The first element corresponds to the innermost circle, and so on, until the last element, that corresponds to the outermost circle. The phaseoffset can be a scalar only if all the elements of M are the same value.

Example: [pi/4 pi/12 pi/16] defines three constellation PSK ring phase offsets. The inner ring has a phase offset of pi/4, the second ring has a phase offset of pi/12, and the outer ring has a phase offset of pi/16.

Data Types: double

Name-Value Arguments

Specify optional pairs of arguments as Name1=Value1,...,NameN=ValueN, where Name is the argument name and Value is the corresponding value. Name-value arguments must appear after other arguments, but the order of the pairs does not matter.

Example: Z = apskdemod(Y,M,radii,OutputType='bit',OutputDataType='single');

SymbolMapping — Symbol mapping

'gray' | 'contourwise-gray' | integer vector

Symbol mapping, specified as one of the following:

• 'contourwise-gray' — Uses Gray mapping along the contour in phase dimension for each PSK ring.

• 'gray' — Uses Gray mapping along the contour in both the amplitude and phase dimensions. For Gray symbol mapping, all the values for M must be equal and all the values for phaseoffset must be equal. For a description of the Gray mapping used, see [2].

• integer vector — Use custom symbol mapping. Vector must consist of sum(M) unique elements with values in the range [0, (sum(M)-1]. The first element corresponds to the constellation point in the first quadrant of the innermost circle, with subsequent elements positioned counterclockwise around the PSK rings.

The default symbol mapping depends on M and phaseoffset. When all the elements of M and phaseOffset are equal, the default is 'gray'. For all other cases, the default is 'contourwise-gray'.

Data Types: double | char | string

OutputType — Output type

'integer' (default) | 'bit' | 'llr' | 'approxllr'

Output type, specified as 'integer', 'bit', 'llr', or 'approxllr'. For a description of the returned output, see Z.

OutputDataType — Output data type

'double' | 'single' | ...

Output data type, specified as one of the data types listed in this table. Acceptable values for OutputDataType depend on the OutputType value.

OutputType Value	Acceptable OutputDataType Values
'integer'	'double', 'single', 'int8', 'int16', 'int32', 'uint8', 'uint16', or 'uint32'
'bit'	'double', 'single', 'int8', 'int16', 'int32', 'uint8', 'uint16', 'uint32', or 'logical'

The default value is the data type of input Y.

Dependencies

To enable this argument, set OutputType to either 'integer' or 'bit'.

NoiseVariance — Noise variance

1 (default) | positive scalar | vector of positive values

Noise variance, specified as one of these options:

• Positive scalar — The function uses the same noise variance value on all input elements.

• Vector of positive values — For all the elements of the input along the corresponding last dimension, the function uses the noise variance specified by each element of the vector. The vector length must be equal to the number of columns in the input signal.

When the noise variance or signal power result in computations involving extreme positive or negative magnitudes, see APSK Soft Demodulation for algorithm selection considerations.

Dependencies

This argument applies when you set OutputType to 'llr' or 'approxllr'.

Data Types: double

PlotConstellation — Option to plot constellation

false or 0 (default) | true or 1

Option to plot constellation, specified as logical 0 (false) or 1 (true). To plot the constellation, set PlotConstellation to true.

Data Types: logical

Output Arguments

Z — Demodulated signal

scalar | vector | matrix

Demodulated signal, returned as a scalar, vector, or matrix. The value and dimension of this output vary depending on the specified OutputType value, as shown in this table.

OutputType Value	apskdemod Output Value	Dimensions of Output
'integer'	Demodulated integer values from 0 to (sum(M) – 1)	Z has the same dimensions as input Y.
'bit'	Demodulated bits	The number of rows in Z is log2(sum(M)) times the number of rows in Y. Each demodulated symbol is mapped to a group of log2(sum(M)) elements in a column, where the first element represents the MSB and the last element represents the LSB.
'llr'	Log-likelihood ratio value for each bit
'approxllr'	Approximate log-likelihood ratio value for each bit

Algorithms

APSK Hard Demodulation

The hard demodulation algorithm applies amplitude phase decoding as described in [1].

APSK Soft Demodulation

For soft demodulation, two soft-decision log-likelihood ratio (LLR) algorithms are available: exact LLR and approximate LLR. The exact LLR algorithm is more accurate but has slower execution speed than the approximate LLR algorithm. For further description of these algorithms, see the Hard- vs. Soft-Decision Demodulation topic.

Note

The exact LLR algorithm computes exponentials using finite precision arithmetic. For computations involving very large positive or negative magnitudes, the exact LLR algorithm yields:

• Inf or -Inf if the noise variance is a very large value

• NaN if the noise variance and signal power are both very small values

The approximate LLR algorithm does not compute exponentials. You can avoid Inf, -Inf, and NaN results by using the approximate LLR algorithm.

References

[1] Sebesta, J. “Efficient Method for APSK Demodulation.” Selected Topics on Applied Mathematics, Circuits, Systems, and Signals (P. Pardalos, N. Mastorakis, V. Mladenov, and Z. Bojkovic, eds.). Vouliagmeni, Athens, Greece: WSEAS Press, 2009.

[2] Liu, Z., Q. Xie, K. Peng, and Z. Yang. "APSK Constellation with Gray Mapping." IEEE Communications Letters. Vol. 15, Number 12, December 2011, pp. 1271–1273.

Extended Capabilities

C/C++ Code Generation
Generate C and C++ code using MATLAB® Coder™.

Version History

Introduced in R2018a